Published on Web 01/24/2002
Investigation of the Mechanism of n-Butane Oxidation on Vanadium Phosphorus Oxide Catalysts: Evidence from Isotopic Labeling Studies Bin Chen and Eric J. Munson* Contribution from the Department of Chemistry, UniVersity of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455 Received February 2, 2001
Abstract: The selective oxidation of n-butane to maleic acid catalyzed by vanadium phosphates (VPO) is one of the most complex partial oxidation reactions used in industry today. Numerous reaction mechanisms have been proposed in the literature, many of which have butenes, butadiene, and furan as reaction intermediates. We have developed an experimental protocol to study the mechanism of this reaction in which 13C-isotopically labeled n-butane is flowed over a catalyst bed and the reaction products are analyzed using 13C NMR spectroscopy. This protocol approximates the conditions found in an industrial reactor without requiring an exorbitant amount of isotopically labeled material. When [1,4-13C]n-butane reacted on VPO catalysts to produce maleic acid and butadiene, the isotopic labels were observed in both the 1,4 and 2,3 positions of butadiene and maleic acid. The ratio of label scrambling was typically 1:20 for the 2,3:1,4 positions in maleic acid. For butadiene, the ratio of label scrambling was consistently much higher, at 2:3 for the 2,3:1,4 positions. Because of the discrepancy in the amount of label scrambling between maleic acid and butadiene, butadiene is unlikely to be the primary reaction intermediate for the conversion of n-butane to maleic anhydride under typical industrial conditions. Ethylene was always observed as a side product for n-butane oxidation on VPO catalysts. Fully 13C-labeled butane produced about 5-13 times as much isotopically labeled ethylene as did [1,4-13C]butane, indicating that ethylene was produced mainly from the two methylene carbons of n-butane. When the reaction was run under conditions which minimize total oxidation products such as CO and CO2, the amounts of ethylene and carbon oxides produced from fully 13C-labeled butane were almost equal. This strongly suggests that the total oxidation of n-butane on VPO catalysts involves the oxidation and abstraction of the two methyl groups of n-butane, and the two methylene groups of n-butane form ethylene. An organometallic mechanism is proposed to explain these results.
Introduction
The selective oxidation of n-butane to maleic anhydride is recognized as one of the most complex selective oxidation reactions used in industry today.1-12 In this reaction, eight hydrogen atoms are abstracted from n-butane, and three oxygen atoms are added to form maleic anhydride. The selectivity to * To whom correspondence should be addressed. (1) Birkeland, K. E.; Babitz, S. M.; Bethke, G. K.; Kung, H. H. J. Phys. Chem. B 1997, 101, 6895-6902. (2) Centi, G.; Trifiro`, F.; Ebner, J. R.; Franchetti, V. M. Chem. ReV. 1988, 88, 55-80. (3) Centi, G. Catal. Today 1993, 16, 5-26. (4) Cavani, F.; Trifiro`, F. Catalysis 1994, 11, 246-315. (5) Abon, M.; Volta, J.-C. Appl. Catal. 1997, 157, 173-193. (6) Hutchings, G. J.; Kiely, C. J.; Sananes-Schulz, M. T.; Burrows, A.; Volta, J. C. Catal. Today 1998, 40, 273-286. (7) Herrmann, J.-M.; Vernoux, P.; Be´re´, K. E.; Abon, M. J. Catal. 1997, 167, 106-117. (8) Hutchings, G. J.; Sananes, M. T.; Sajip, S.; Kiely, C. J.; Burrows, A.; Ellison, I. J.; Volta, J. C. Catal. Today 1997, 33, 161-171. (9) Aı¨t-Lachgar, K.; Tuel, A.; Brun, M.; Herrmann, J. M.; Krafft, J. M.; Martin, J. R.; Volta, J. C.; Abon, M. J. Catal. 1998, 177, 224-230. (10) Hutchings, G. J.; Desmartln-Chomel, A.; Oller, R.; Volta, J. C. Nature 1994, 368, 41-45. (11) Gai, P. L.; Kourtakis, K. Science 1995, 267, 661-663. (12) Coulston, G. W.; Bare, S. R.; Kung, H.; Birkeland, K.; Bethke, G. K.; Harlow, R.; Herron, N.; Lee, P. L. Science 1997, 275, 191-193. 1638 VOL. 124, NO. 8, 2002
9
J. AM. CHEM. SOC.
maleic anhydride is around 65-70% under typical industrial conditions (less than 2 molar % n-butane in air, with conversion from 70 to 85% between 400 and 450 °C).2 Only maleic anhydride (or acid), carbon oxides, and a trace amount of acetic acid are detected under such conditions. All catalysts used industrially for the production of maleic anhydride from n-butane are based on vanadium phosphorus oxides (VPO).2 Vanadyl pyrophosphate ((VO)2P2O7) is regarded to be the active phase of the VPO catalysts since it is the single crystalline phase that exists in equilibrated VPO catalysts (after more than 200 h on stream).2 The average oxidation state of vanadium in equilibrated VPO catalysts is typically 4.00-4.03.2 Despite extensive study, little is known about the reaction mechanism for n-butane oxidation on VPO catalysts. One reason is that none of the potential reaction intermediates, such as butenes, butadiene, furan, etc., have been observed in the products using typical industrial conditions.13 A common approach has been to run the reaction under unusual reaction conditions to detect possible reaction intermediates. Centi et (13) Centi, G.; Fornassari, G.; Trifiro`, F. Ind. Eng. Chem. Prod. Res. DeV. 1985, 24, 32-37. 10.1021/ja010285v CCC: $22.00 © 2002 American Chemical Society
n-Butane Oxidation on VPO Catalysts
al.14 have shown that at high butane concentrations, butenes, butadiene, and furan, but no maleic anhydride, were detected using a traditional flow reactor. Another approach has been nonequilibrium transient experiments which employ a temporal analysis of products (TAP) reactor developed by Gleaves et al.15-17 The reactor consisted of high-speed injection valves, a microreactor, and a quadrupole mass spectrometer separated from the reactor by differentially pumped chambers.15 Butenes, butadiene, and furan were detected sequentially with respect to their maximum intensity on equilibrated (VO)2P2O7 catalyst. However, when the catalyst was first pulsed extensively with oxygen at reaction temperature (420 °C) and subsequently pulsed with n-butane, maleic anhydride was observed, but the proposed intermediate products were not.15 The following reaction pathway was proposed on the basis of these results:4,14
However, several arguments have been made against butenes, butadiene, and furan as reaction intermediates:2,4 1. These compounds were detected under very unusual conditions, such as low oxygen and very high n-butane concentrations and at very low contact times, or under high vacuum in the TAP reactor, or in the oxidation of n-butane under anaerobic conditions in a pulse reactor. 2. There was no desorption of intermediates during n-butane oxidation in the presence of available oxygen (either molecular oxygen or lattice oxygen associated with V5+). 3. The oxidation of n-butane and the intermediate compounds on VPO catalysts yielded different product distributions. As mentioned above, only maleic anhydride, carbon oxides, and a trace amount of acetic acid were detected for n-butane oxidation. By contrast, acetaldehyde, crotonaldehyde, and other partial oxidation products were detected in the case of the oxidation of C4 olefins.2 Kinetic methods have also been used to study the reaction mechanism. Zhang-Lin et al.18 conducted kinetics studies of the oxidation of n-butane, butadiene, furan, and maleic anhydride on various VPO phases to investigate the mechanism of n-butane oxidation on VPO catalysts. They concluded that the main route from butane to maleic anhydride is an “alkoxide route” in which the precursors to maleic anhydride are alkoxide species. These alkoxides maintain a σ-bond between the substrate and the catalyst surface, and there is no desorption in the gas phase of butenes, butadiene, and furan. By using crystallochemical models of active sites and examining the energetics and geometries of butane oxidation on the (100) face of (VO2)2P2O7, Ziolkowski et al.19 reported that the active site for the direct oxidation of butane to maleic anhydride is situated between four protruding, undersaturated oxygens (2 × V-O, 2 × P-O). The activation of butane consists of the abstraction of a H from each (14) Centi, G.; Fornasari, G.; Trifiro`, F. J. Catal. 1984, 89, 44-51. (15) Gleaves, J. T.; Ebner, J. R.; Kuechler, T. C. Catal. ReV. Sci. Eng. 1988, 30, 49-116. (16) Rodemerck, U.; Kubias, B.; Zanthoff, H.-W.; Baerms, M. Appl. Catal. A: General 1997, 153, 203-216. Rodemerck, U.; Kubias, B.; Zanthoff, H.W.; Baerms, M. Appl. Catal. A: General 1997, 153, 217-231. (17) Kubias, B.; Rodemerck, U.; Zanthoff, H.-W.; Meisel, M. Catal. Today 1996, 32, 243-253. (18) Zhang-Lin, Y.; Forissier, M.; Sneeden, R. P.; Ve´drine, J. C.; Volta, J. C. J. Catal. 1994, 145, 256-266. Zhang-Lin, Y.; Forissier, M.; Ve´drine, J. C.; Volta, J. C. J. Catal. 1994, 145, 267-275. (19) Ziolkowski, J.; Bordes, E.; Courtine, P. J. Catal. 1990, 122, 126-150.
ARTICLES
methyl group, with the concerted formation of two strong Cterminal-Osurf bonds so that the molecule is anchored long enough for the reaction to be completed. However, experiments with deuterium-labeled n-butane revealed that the first step of butane oxidation is the irreversible activation of a methylene C-H bond in butane on the catalyst surface.20 In situ Fourier transform infrared (FTIR) spectroscopy studies have been performed to investigate butane oxidation on VPO catalysts.21,22 Wenig and Schrader used an in situ FT-IR cell to study the interaction of n-butane with VPO catalysts.22 These authors reported evidence for the presence of n-butane, maleic anhydride, carbon oxides, and reactive surface species (maleic acid and olefins) on the catalyst at temperatures of 200-400 °C. Recently, Xue and Schrader developed a technique called “transient FTIR” which uses special operation techniques such as pulse reaction and reactant feed cycling to observe the evolution of the IR spectra as a function of time.23 Results from transient FTIR studies suggested that unsaturated noncyclic carbonyl species may be precursors to maleic anhydride and butane might be adsorbed on the VPO catalyst to form olefinic species at low temperatures (50 °C). The latter point contradicts general observations that the activation of butane to butenes is the rate-limiting step and occurs at much higher temperature (>350 °C).2 Recently, we have developed an experimental protocol which utilizes selective 13C isotopic labeling and examination of the reaction products by 13C NMR spectroscopy to investigate the mechanism of n-butane oxidation on VPO catalysts. The advantage of this protocol is that the fate of the 13C label can be monitored after the reaction and therefore gives insights into the reaction mechanism. Previously we have shown that the label in butadiene produced from [1,4-13C]n-butane is completely scrambled, but in maleic acid, also produced from [1,4-13C]nbutane, the label is largely unscrambled. This makes it unlikely that maleic acid is formed predominantly by a butadiene intermediate.24 In this paper, we describe our experimental results for all aspects of this reaction obtained using this protocol. We have found that ethylene was always a side product for the reaction of n-butane on VPO catalysts. When fully 13C-labeled butane was used instead of [1,4-13C]butane, the 13C NMR peak intensity of ethylene increased by 5-13 times, showing that ethylene was produced mainly from the two methylene carbons of n-butane. Moreover, the yields of carbon oxides and ethylene were roughly equal on catalysts with phosphorus:vanadium ratios slightly higher than the stoichiometric ratio (P:V ) 1.2 and 1.1). These catalysts had higher selectivities for maleic acid production. This result suggests that the total oxidation of n-butane on the selective VPO catalysts involves mainly the oxidation and abstraction of the two methyl carbons, and the two methylene groups are left to form ethylene. The ratio of label scrambling in maleic acid changed significantly when n-butane reacted on the VPO catalysts with higher P:V ratios. When the P:V ratio changed from 0.9 to 1.2, the ratio of label scrambling changed from 1:30 to 1:6 for the 2,3:1,4 positions (20) Pepera, M. A.; Callahan, J. L.; Desmond, M. J.; Milberger, E. C.; Blum, P. R.; Bremer, N. J. J. Am. Chem. Soc. 1985, 107, 4883-4892. (21) Busca, G. Catal. Today 1996, 457-496. (22) Wenig, R. W.; Schrader, G. L. J. Phys. Chem. 1987, 91, 5674-5680. (23) Xue, Z.-Y.; Schrader, G. L. J. Catal. 1999, 184, 87-104. (24) Chen, B.; Munson, E. J. J. Am. Chem. Soc. 1999, 121, 11024-11025. J. AM. CHEM. SOC.
9
VOL. 124, NO. 8, 2002 1639
ARTICLES
Chen and Munson
in maleic acid. Concurrently, more ethylene was observed when [1,4-13C]butane reacted on the VPO catalysts with higher P:V ratios. The correlation between the ratio of label scrambling and the peak intensity of ethylene when [1,4-13C]butane was the reactant indicates that there might be an association between the formation of ethylene and the occurrence of label scrambling. There was no ethylene formed and no label scrambling was observed when butadiene reacted on the VPO catalysts to produce maleic acid. From these results we are able to propose a plausible mechanism for this reaction. We propose that n-butane is absorbed on the VPO catalysts via an organometallic interaction between the two methyl carbons of butane and the vanadium atom on the surface of the catalyst during the reaction, thus forming a chelating configuration. The chelating configuration serves to hold the butane molecule on the surface of the catalyst while hydrogens of butane are removed from it and oxygens are added to it until maleic anhydride is formed. The chelating configuration of the reaction intermediates is used to explain label scrambling, formation of ethylene, and other observations. Experimental Section Synthesis of VPO Catalysts. The catalysts used in this study can be divided into two groups: catalysts with constant P:V ratio (1:1) but different average vanadium oxidation states, and catalysts with different P:V ratios. Catalysts are denoted either by their average vanadium oxidation states, e.g., VPO3.92, or the P:V ratio, e.g., P:V ) 0.9. The procedures for making these catalysts are described as follows. (i) Catalysts with Constant P:V Ratio (1:1). Precursor 1. Stoichiometric amounts (P:V ) 1:1) of V2O5 and H3PO4 (85%) were mixed in ethanol and refluxed for 16 h. The resulting precursor was blue and identified by XRD as VOHPO4‚0.5H2O. VPO3.92, VPO4.75, and VPO4.95 were made from precursor 1:
Figure 1. Pseudo-flow reactor.
(ii) Catalysts with Different P:V Ratios (0.9-1.2). Catalysts with various P:V ratios were made using the following procedure: 5 g of V2O5 and a stoichiometric amount of H3PO4 (85%) (P:V ) 0.9-1.2) were mixed in 100 g of isobutanol and refluxed for 24 h with continuous
stirring. The suspension was then transferred to a 1000-mL beaker and heated gently on a hot plate to near dryness. The residue was then dried at 110 °C for 24 h. The resulting precursor was calcined at 380 °C for 3 h in air to obtain the catalyst. Characterization of the Catalysts. The average valence state of vanadium in the VPO catalysts was determined using the potentiometric method described by Niwa and Murakami.26 Specific surface areas of the VPO catalysts were measured using the BET method (Micromeritics ASAP 2000). The XRD patterns of the catalysts were obtained using a X-ray diffractometer (SIEMENS D5005). Pseudo-Flow Reaction. A pseudo-flow reactor was designed to prepare samples for NMR analysis (Figure 1). VPO catalyst (∼0.2 g) was placed in the center of the glass reactor. The reactor was attached to a high-vacuum line and evacuated to 15), total oxidation of n-butane dominated the reaction at the beginning. Total oxidation of n-butane significantly decreased the concentration of V5+ on the surface of the catalyst, which favored label scrambling in the subsequent selective oxidation of n-butane to maleic acid. Again, ethylene and methanediol were always detected. Methanediol became a major partial oxidation product at oxygen/butane ratios of 20:1 and 22:1. The drawback of the above-mentioned approaches is that oxygen and n-butane react sequentially with the catalyst, which is not exactly the case with industrial reactions in fixed-bed flow reactors. Another experiment was performed to better mimic true industrial conditions. A vacuum valve was added to the pseudo-flow reactor between the reactant chamber and the catalyst. Oxygen and [1,4-13C]butane (O2/butane ) 10) were mixed before the valve was opened. The gaseous mixture was then allowed to flow through VPO3.92 at 380 °C. Shown in Figure 6 is the 13C MAS NMR spectrum of the reaction products. The formation of maleic acid, ethylene, and meth1644 J. AM. CHEM. SOC.
9
VOL. 124, NO. 8, 2002
anediol and label scrambling were consistent with previous results, except that the percent conversion of the reaction was much higher, showing that the presence of gaseous oxygen during the reaction enhanced the activity of the catalyst. Effect of Oxidation State of the Catalyst on Label Scrambling. The most unexpected result that we observed was 13C label scrambling in maleic acid. While the amount of labelscrambled maleic acid might be very low, it could provide valuable insight into the mechanism of the reaction. To further understand the mechanistic significance of label scrambling, the effects of various factors, such as oxidation state and P:V ratio of the catalyst, reaction temperature, etc., on label scrambling were investigated. The percentage of label exchange for maleic acid varied between 1% and 5% when the initial oxidation state of vanadium changed between 4.34 and 4.95. Because these catalysts were made from different precursors and showed different activities, it is hard to estimate the effect of the oxidation state of vanadium on label scrambling using the data from these catalysts. To simplify the situation, sequential reactions of [1,4-13C]butane on VPO4.95 were performed to investigate the effect of reduction of the catalyst on label scrambling. Five sequential reactions were performed on VPO4.95 at 380 °C, and the ratio of label scrambling changed from 1:100 to 1:20 (2,3:1,4 in maleic acid) as the catalyst was reduced, showing that the lower oxidation state of the catalyst favors label scrambling. Effect of P:V Ratio on Label Scrambling. The phosphorusto-vanadium (P:V) ratio is a key parameter in determining the selectivity and activity of VPO catalysts.2-4,8,28,29 A slight excess of phosphorus with respect to the stoichiometric amount (an (28) Wenig, R. W.; Schrader, G. L. Ind. Eng. Chem. Fundam. 1986, 25, 612620. Wenig, R. W.; Schrader, G. L. J. Phys. Chem. 1986, 90, 6480-6488. (29) Hannour, F. K.; Martin, A.; Kubias, B.; Lu¨cke, B.; Bordes, E.; Courtine, P. Catal. Today 1998, 40, 263-272.
n-Butane Oxidation on VPO Catalysts
Figure 7. 13C MAS NMR spectra of the reaction products of [1,4-13C]butane on the VPO catalysts with different P:V ratios. The average oxidation state of vanadium in these catalysts was 4.47-4.97 (Table 1). The reaction temperature was 380 °C.
atomic P:V ratio in the range of 1.01-1.10) was necessary to obtain an optimal catalyst.4 For this reason the effect of P:V ratio on label scrambling was investigated. [1,4-13C]Butane was allowed to react at 380 °C on VPO catalysts with P:V ratios ranging from 0.9 to 1.2. 13C MAS NMR spectra of the products of these reactions are shown in Figure 7. The ratio of label scrambling in maleic acid and fumaric acid changed consistently as the P:V ratio changed from 1.2 to 0.9. When P:V ) 1.2, the ratio of label scrambling was estimated to be 1:6 (2,3:1,4 in maleic acid and fumaric acid); when P:V ) 0.9, the peaks from the label scrambled carbons were very small, though the signal/ noise ratio does not allow an accurate estimation of the ratio of label scrambling. The precursors of the catalysts with different P/V ratios (0.9-1.2) have essentially the same oxidation state (∼4.0). However, large variations (4.47-4.97) in the oxidation state of the catalysts were obtained after the precursors were calcined under identical conditions (Table 1). It has been shown that higher oxidation states result in less label scrambling, which makes it difficult to attribute changes in the ratio of label scrambling to P:V ratio only. When [1,4-13C]butane reacted at 550 °C on the precursors of the catalysts with different P:V ratios to produce butadiene, the percentage of label-scrambled butadiene was consistently 40-45% (Figure 8). This result is consistent with our previous experiments on VPO3.92. Formation of Carbon Oxides and Ethylene. Previous experiments showed that ethylene is always a byproduct in the oxidation of n-butane on the VPO catalysts. Figure 9 shows the 13C MAS NMR spectra of the reaction products of fully 13C-labeled butane and [1,4-13C]butane on VPO catalysts with different P:V ratios. The ratios of the integrated peak intensities of ethylene and carbon oxides (CO and CO2) from either fully 13C-labeled butane or [1,4-13C]butane were calculated and are shown in Figure 10. The peak intensity of ethylene increased
ARTICLES
Figure 8. 13C MAS NMR spectra of the reaction products of [1,4-13C]butane on the precursors of the VPO catalysts with different P:V ratios. The reaction temperature was 550 °C.
Figure 9. 13C MAS NMR peaks of maleic acid (and fumaric acid), carbon dioxide, and ethylene produced from the reaction of [1,4-13C]butane and fully 13C-labeled butane on VPO catalysts with different P:V ratios at 380 °C: (A) fully 13C-labeled butane and (B) [1,4-13C]butane.
by a factor of 5-13 when fully 13C-labeled butane was used instead of [1,4-13C]butane. A similar result has been observed on VPO4.56 (Figure 3b,f) and other VPO catalysts. This result suggests that ethylene was likely produced mainly from the two J. AM. CHEM. SOC.
9
VOL. 124, NO. 8, 2002 1645
ARTICLES
Chen and Munson
Figure 11. Selectivity to maleic and fumaric acids and conversion of n-butane on VPO catalysts with different P:V ratios. The reaction temperature was 380 °C. Figure 10. Ratios of the integrated peak intensity of COx (sum of CO and CO2) and ethylene produced from the reaction of either [1,4-13C]butane or fully 13C-labeled butane on the VPO catalysts with different P:V ratios. The reaction temperature was 380 °C. COx and C2H4 correspond to carbon oxides and ethylene produced from [1,4-13C]butane. COx* and C2H4* correspond to carbon oxides and ethylene produced from fully 13C-labeled butane.
methylene carbons of n-butane, which was not observed in the 13C NMR spectra when [1,4-13C]butane was the reactant. By further examining the 13C MAS NMR spectra for the oxidation of fully 13C-labeled butane (Figure 9), it was found that the ratio of peak intensity of ethylene to carbon oxides (sum of CO2 and CO) is close to 1 for P:V ) 1.2 and 1.1. This result suggests that there might be a relationship between the formation of ethylene and the formation of carbon oxides on the catalysts with P:V ) 1.2 and 1.1. Also, the peak intensity of carbon oxides did not change significantly (fully 13C-labeled butane versus [1,4-13C]butane) for the catalysts with P:V ratios of 1.2 and 1.1 (Figure 10), suggesting that the total oxidation of n-butane involves mainly the two methyl groups on these two catalysts. The combination of these results suggests that the formation of ethylene and carbon oxides might be associated with each other via the following mechanism on the catalysts with P:V ratios of 1.2 and 1.1:
The total oxidation products (CO and CO2) were formed primarily from the two methyl groups of butane, and the two methylene groups formed ethylene. The P:V ratio plays a key role in determining the selectivity and activity of the VPO catalysts. Catalysts with P:V slightly greater than 1 had a much higher selectivity and much lower activity than the catalysts with P:V ) 1 or P:V < 1 (Figure 11). Our results suggest that the mechanism described in eq 8 might be the major mechanism for the total oxidation of n-butane on the more selective VPO catalysts. Equation 8 is not the only mechanism for the formation of ethylene, because ethylene was always detected in the 13C NMR spectra when [1,4-13C]butane was the reactant. The observed ethylene in the case of [1,4-13C]butane oxidation was likely produced from the following mechanism:
1646 J. AM. CHEM. SOC.
9
VOL. 124, NO. 8, 2002
The cleavage of the C2-C3 bond of butane resulted in the formation of ethylene from either C1-C2 or C3-C4 of butane. This is the minor mechanism for the formation of ethylene. Ethylene formation by this mechanism might be related to the mechanism of label scrambling, because more ethylene was observed at higher ratios of label scrambling. Our results also suggest that there is more than one mechanism for the total oxidation of butane, especially on the nonselective catalysts (P:V ) 0.90-1.0). Fully 13C-labeled butane yielded more carbon oxides than [1,4-13C]butane on the catalysts with lower P:V ratio (Figure 10), which were also less selective (Figure 11). When P:V ) 0.90, which was the least selective catalyst, the peak intensity of carbon oxides almost doubled for fully 13C-labeled butane compared to those for [1,413C]butane, indicating that the predominant mechanism for total oxidation of butane on this catalyst was the conversion of the entire butane molecule to carbon oxides:
Further evidence is that the ratio of peak intensity of ethylene/ COx decreased with P:V ratio and dropped to
